60 study points
Studies of Electron Transfer from Flavodoxin Reductase to NrdI in Bacillus cereus
by Silje Skråmo
Department of Molecular Biosciences Faculty of Mathematics and Natural Biosciences
UNIVERSITY OF OSLO 06/2012
The work presented in this thesis has been carried out in the laboratory of Professor K.
Kristoffer Andersson at the Department of Molecular Biosciences, University of Oslo.
First, I want to thank Professor K. Kristoffer Andersson for letting me do a master project in his group, and for introducing me to the interesting field of metalloproteins that they are working in. I am also very thankful for the conferences, meetings and synchrotron trips he has sent me to. In addition to him, have my co-supervisors Hans-Petter Hersleth, Marta
Hammerstad and Åsmund K. Røhr given me excellent supervision during the whole period that I have been working on my master project, and we have had interesting talks and discussions that have taught me a lot. I want to thank them for that, and for always being there.
I will also thank Marie, Mari, my other fellow students and the group for the great social environment they make; for the lunch breaks, the laughter, the dinners, the trips, and for always making me smile.
Finally, I want to thank Bjørn, mamma, pappa and Kim for all the support, love and understanding.
Oslo, June 2012 Silje Skråmo
Ribonucleotide reductase (RNR) catalyses the reduction of the four ribonucleotides to their corresponding deoxyribonucleotides needed for synthesis and repair of DNA. RNR is found in the genome of all living organisms, indicating that this enzyme is an essential part of the fundamentals of life. The RNRs explored to date are divided into three different main classes;
Ia-c, II and III.
The cofactor of class Ib RNRs was recently shown to be a dimanganese cluster. Further that the flavoprotein NrdI is essential for the generation of the dimanganese-tyrosyl radical, the active form of the cofactor. An active cofactor is, in turn, essential for generation of a protein radical in the catalytic site, which initiates substrate turnover. The function of NrdI in class Ib RNR is, however, not fully understood. Two flavodoxin reductases (FldRs) have been located in the genome of Bacillus cereus (B. cereus), and considered as potential reducers of NrdI by transferring electrons from NADPH to NrdI. Investigations in their ability to interact with and reduce NrdI, would enhance the knowledge about the function of NrdI.
This master project has focused on the two FldRs found in B. cereus. One of them was cloned, and both were expressed and purified. One of the FldRs is expressed primary as apo protein, and work was performed to reconstitute cofactor-containing protein from apo-FldR and flavin. Initial protein crystals were obtained for both proteins, but no complete datasets could be collected. The ability of each FldR to reduce NrdI and two other flavodoxins from B.
cereus was tested. The result suggests that the redox partners of both FldRs are other than NrdI. Initial kinetic measurements were performed with 2,6-dichlorophenolindophenol (DCPIP) as the electron acceptor, resulting in a 200-fold difference in the Km values of the respective FldRs. Their ability to reduce cytochrome c (Cytc) was also tested. In sum, the experiments with the two-electron acceptor DCPIP and the one-electron acceptor Cytc indicated that one FldR is the most effective two-electron donor, and the other FldR is the most effective one-electron donor.
1. INTRODUCTION...1
1.1. Ribonucleotide reductase.......1
1.1.1. Many different ribonucleotide reductases...1
1.1.2. The enzymatic reaction...4
1.2. Class Ib RNR...6
1.2.1. The identity of the class Ib RNR metal cofactor...6
1.2.2. Generation of an active metal cofactor – the importance of NrdH and NrdI...7
1.3. Aims of the study...12
1.4. Protein X-ray crystallography...14
1.4.1. Protein X-ray crystallography...14
1.4.2. Scattering...15
1.4.3. The nature of protein crystals and crystal growth...15
1.4.4. Vapour diffusion...16
2. METHODS...17
2.1. Cloning of BC0385 into the pET-22b expression vector and generation of FldR BC0385 expressing E. coli cells...17
2.1.1. Amplification of BC0385...17
2.1.2. Purification of pET-22b vector...18
2.1.3. Restriction cutting and ligation of BC0385 and pET-22b vector...19
2.1.4. Transformation of competent hosts with the ligation product...20
2.1.5. Site-directed mutagenesis...20
2.1.6. Transformation of expression hosts with the plasmid...21
2.2. Expression and purification of FldR BC0385 and FldR BC4926...22
2.2.1. Growth of FldR expressing E. coli cells...22
2.2.2. Testing the expression of FldR...23
2.2.5. Purification of FldR: Protein precipitation by salting out...24
2.2.6. Purification of FldR: Desalting chromatography...25
2.2.7. Purification of FldR: anion exchange chromatography...25
2.2.9. Purification of FldR: Gel filtration chromatography...26
2.3. Crystallization of FldR BC0385 and FldR BC4926...27
2.3.1. Screening for crystallisation conditions...27
2.3.2. Optimisation of crystallisation conditions for apo-FldR BC4926...28
2.3.3. Optimisation of crystallization conditions for FldR BC4926...29
2.4. X-ray diffraction, data collection and data processing...29
2.5. Activity measurements...30
2.5.1. Determination of the specific activity of the FldRs with DCPIP...30
2.5.2. Comparison of the FldR activities with cytochrome c...31
2.5.3. Testing of the ability of the FldRs to reduce flavodoxins from B. cereus...31
3. RESULTS AND DISCUSSION...33
3.1. Cloning of BC0385 into the pET-22b expression vector and generation of FldR BC0385 expressing E. coli cells...33
3.2. Expression and purification of FldR BC0385 and FldR BC4926...36
3.2.1. Purification of FldR BC0385...37
3.2.2. Purification of FldR BC4926...39
3.3. Protein crystallisation and X-ray diffraction...44
3.3.1. Crystallisation of FldR BC0385...44
3.3.2. X-ray diffraction experiment on the FldR BC0385 crystal...45
3.3.3. Crystallisation of FldR BC4926...45
3.3.4. X-ray diffraction experiment on FldR BC4926 crystals...47
3.4. Activity measurements...47
3.4.3. Testing of the ability of the FldRs to reduce flavodoxins from B. cereus...51 4. CONCLUDING DISCUSSION AND FURTHER PERSPECTIVES...53 REFERENCES...57 APPENDIX 1: Materials...I APPENDIX 2: PCR mixtures and PCR programs...V APPENDIX 3: Restriction- and ligation reaction mixtures...VI APPENDIX 4: Media, buffers, gels and reaction mixtures...VII APPENDIX 5: Nucleic acid sequences and primer sequences...X
B. cereus Bacillus cereus
BL21-cells E. coli BL21 (DE3) Competent cells
Cytc cytochrome c
DCPIP 2,6-dichlorophenolindophenol
E. coli Escherichia coli
Fld flavodoxin
FldR Flavodoxin reductase
FMN flavin mononucleotide
hq hydroquinone
LB Luria Broth
NADH nicotinamide adenine dinucleotide
NADPH nicotinamide adenine dinucleotide
ox oxidised
PCR polymerase chain reaction
PX protein X-ray crystallography
redox reduction-oxidation
RNR Ribonucleotide reductase
SDS-PAGE sodium dodecyl sulphate polyacrylamide gel electrophoresis
sq semiquinone
TB Terrific Broth
UV-vis Ultraviolet and visible light
V0 initial velocity
XL10-cells E. coli XL10-Gold ultracompentent cells
Figure 1.1: Ribonucleotide di- and triphosphates are converted to deoxyribonucleotide di- and triphosphates by reduction of the 2’-hydroxyl group in the ribose ring.
1. Introduction
1.1. Ribonucleotide reductase
1.1.1. Many different ribonucleotide reductases
Deoxyribonucleotides, the precursors of DNA, are generated by the reduction of
ribonucleotides, the precursors of RNA. The reduction is a radical-induced process, and takes place by the 2′-hydroxyl group on the ribose ring being replaced by a hydrogen atom (Figure 1.1).
The enzyme ribonucleotide reductase (RNR) catalyses this reaction for all four DNA
precursors, and appear to do so in all living species which so far have been sequenced, as well as in some virus. Hence, it is assumed that all organisms depend on this machinery for de novo production of deoxyribonucleotides. All known RNRs share a common reaction
mechanism and contain two distinct functional components; a ribonucleotide binding site and a radical generator (cofactor). Despite little sequence similarity, detailed studies have revealed that several crucial residues are conserved among the RNRs of different organisms, many of which are located in the substrate binding site. Additionally, the three-dimensional structural features of the RNRs studied to date, indicate that the substrate binding sites of these enzymes are structurally homologues [Berg, Tymoczko and Stryer, 2007; Eklund et al., 2001; Jordan and Reichard, 1998; Nordlund and Reichard, 2006; Tomter et al., 2012].
The first RNR to be discovered some 50 years ago was from Escherichia coli (E. coli) [Reichard, Baldesten and Rutberg, 1961]. Soon after the discovery of the same enzyme in Lactobacillus leichmannii [Blakley and Barker, 1964], it became clear that RNR can catalyse ribonucleotide reduction by the use of different cofactors. Since then, several other RNRs
Figure 1.2: Overview of the different classes of RNR. Adapted from Tomter et al., 2012.
with varying cofactors have been discovered, and are today divided into three main classes, namely Ia-c, II and III [Cotruvo and Stubbe, 2011b] (Figure 1.2). Several organisms encode RNRs from different classes – some even from all three classes – and the expression of the different RNRs depend on the environmental conditions. With the exception of Euglena gracilis [Hamilton, 1973], class Ia RNRs are found in all eukaryotes from yeast to human and in prokaryotes, viruses and bacteriophages. Class Ia RNRs depend on oxygen for functionality, hence active enzymes from this class are not found in strict anaerobes. Class Ib RNRs are found in a wide range of facultative and obligate aerobic prokaryotes. Class Ic was discovered more recently in Chlamydia trachomatis [Roshick, Iliffe-Lee and McClarty, 2000], and determination of its sequence suggests its presence in the genomes of some archaea and eubacteria, with some of the eubacteria being thermophiles [Högbom et al., 2004]. Class II RNRs are microbial enzymes that occur in both aerobic and anaerobic organisms. Class III RNRs depend on anaerobic environments to function, and are found in both strict anaerobic and facultative anaerobic organisms (i.e. organisms that performs aerobic respiration when O2
is present and fermentation when it is absent) and a few unicellular eukaryotes [Cotruvo and Stubbe, 2011b; Jiang et al., 2007; Nordlund and Reichard, 2006; Reichard, 1993].
Figure 1.3: Metal cofactor structures found in class Ia, class Ib and class Ic RNR. A) Class Ia RNR from mouse (PDB code 1W68) with both metal centres occupied by iron, B) class Ib RNR from C. ammoniagenes (PDB code 3MJO) with both metal centres occupied by manganese and C) class Ic RNR from C. trachomatis (coordinates is from Martin Högbom) with one metal centre occupied by iron and the other by manganese. Figure adapted from Tomter et al., 2012.
All three classes are proposed to use a thiyl radical (S•) for the initiation of substrate turnover, and the generation of the S• is accomplished via oxidative power derived from the different cofactors (Figure 1.2). Class Ia RNRs utilises a diiron tyrosyl radical (FeIIIFeIII-Y•) as the cofactor [Sjöberg, et al., 1977], class Ib is more recently been proposed to utilise a
dimanganese tyrosyl radical (MnIIIMnIII-Y•) [Cox et al., 2010; Jordan et al., 1994] instead of a FeIIIFeIII-Y•, and for class Ic it is proposed to be a stable diiron (FeIVFeIII) or manganese-iron (MnIVFeIII) cluster (Figure 1.3) [Jiang et al., 2007, Voevodskaya et al., 2007].
The RNRs of class I are composed of two types of subunits, designated R1 (α) and R2 (β), where R1 housed the substrate binding site and R2 harbours the metal cofactor (Figure 1.2).
The quarternary structures of the class I enzymes are actively investigated, but are best understood for the class Ia RNRs from E. coli and mouse. In this class, there is a general consensus that the active form is α2β2 (Figure 1.4), whereas it is proposed to be more complex in eukaryotes [Fairman et al., 2011; Hofer et al., 2012; Kashlan et al., 2002;
Figure 1.4: The proposed α2β2 model from E. coli, where the subunits are docked along their symmetry axis (PDB codes: 3R1R, 1RIB). Adapted from Tomter et al., 2012.
Rofougaran, Vodnala and Hofer, 2006]. Class II RNRs are O2-independent and use adenosylcobalamin as the S• generator, and class III RNRs use a glycyl radical for this purpose, which is generated by a (4Fe4S)1+/2+ cluster together with S-adenosylmethionine.
These enzymes have structures similar to the R1 subunits of class I, but they lack analogues to the R2 subunits as the nature of the cofactors to a large extent deviates from those of the class I RNRs [Stubbe and van der Donk, 1998] (Figure 1.2).
1.1.2. The enzymatic reaction
The enzymatic reaction was first postulated after experiments with isotopic labelled nucleosides in rat [Hammarsten, Reichard and Saluste, 1950]. 26 years later, a tyrosine residue serving as the protein radical was observed in the polypeptide chain of an RNR, supporting the postulation [Reichard and Ehrenberg, 1983]. The S• of the protein generates a substrate radical, which in turn leads to the reduction of the ribonucleotide to a
deoxyribonucleotide, through a mechanism which is common for all RNRs. The following description focuses on class I (and II) RNRs, but still describes the general mechanism of class II and III RNR as well. Initially, the cofactor of the R2 subunit is oxidised to a protein radical. The radical is transferred ̴ 35 Å to the R1 subunit, where it leads to the oxidation of a conserved cysteine residue in the active site to an S•. The S• then initiates a complex, radical- mediated reduction process by abstracting the nucleotide´s 3′-hydrogen atom, which is returned to the 3′-position at the end of the reaction (Figure 1.5). The direct source of electrons required for nucleotide reduction is two conserved cysteines in the active sites of class I and II RNRs. The re-reduction of this electron source is necessary for multiple turns of
Figure 1.5: The proposed mechanism of ribonucleotide reduction by RNRs. The active sites of all three classes of RNRs share a conserved cysteine on the top face of the substrate. This residue is oxidised to an S• by the radical generating component of the RNR in the first step of catalysis. The S•
initiates substrate reduction by abstraction of the nucleotides’ 3′ hydrogen atom (red), which is returned to the 3′ position in the product at the end of the reaction [Licht, Gerfen and Stubbe, 1996].
nucleotide reduction to take place, and this is accomplished by using electrons from
thioredoxin or glutaredoxin (Figure 1.2) [Cotruvo and Stubbe, 2011b; Kolberg et al., 2004;
Licht, Gerfen and Stubbe, 1996; Mao et al., 1992; Stubbe and van der Donk, 1998]. In class III RNRs, only one of these two cysteines is intact; instead, an asparagine residue is located in an equivalent position. Hence, formation of the disulphide bond, a critical step in the proposed mechanism for class I and II, is clearly not relevant for this class. It has been shown that formate functions as a reductant in class III RNRs (Figure 1.2) [Mulliez, et al., 1995], and on the basis of structural studies, it seems likely that the formate binds in the active site, where it would be directly involved in the reduction reactions [Logan et al., 1999].
1.2. Class Ib RNR
1.2.1. The identity of the class Ib RNR metal cofactor
The identity of the metal ions in the cofactor required for an active class Ib RNR in vivo, has been debated for nearly three decades. However, until more recently it was generally thought that the class Ib RNR contained a FeIIIFeIII-Y• cofactor in the same way as class Ia RNR.
Indications of the existence of a manganese-dependent RNR came from observations in Corynebacterium species. During industrial production of amino acids and nucleotides, it was experienced several defects under manganese deficiency, namely unbalanced growth, cell elongation, loss of viability and arrest of DNA synthesis. At the same time, RNA, protein and peptidoglycan synthesis were unaffected. These indications led to a study of Corynebacterium ammoniagenes (C. ammoniagenes) with regard to the metal ions required for an active RNR [Willing, Follman and Auling, 1978]. The study concluded that manganese ions is most likely required for the formation of an active cofactor rather than iron ions, because the R2-subunit was associated with the radioactive 54Mn from the growth medium [Willing, Follman and Auling, 1978]. The crystal structure of the R2-subunit of RNR from C. ammoniagenes was reported many years later, with the observation that the metal site contains an oxo/hydroxo- bridged manganese dimer, located near a tyrosine residue, hence confirming the previous indications [Cox et al., 2010]. Moreover, recent quantification of the metal content in class Ib RNR from another species of this genus, namely Corynebacterium glutamicum, revealed 0.06 mol Fe but 0.8 mol Mn per mol R2-monomer, which would assign two manganese atoms bound to the dimeric metal cofactor [Abbouni et al., 2009]. The fact that E. coli class Ib RNR is expressed under conditions of iron-limitation and oxidative stress, has also supported the existence of a class I RNR that requires a metal other than iron to obtain an active RNR [Monje-Casas et al., 2001; McHugh et al., 2003]. Although class Ib RNR from E. coli have shown to be active with a FeIIIFeIII-Y• cofactor in vitro [Roca et al., 2008], this enzyme is also active in vitro in its manganese form, i.e. with a MnIIIMnIII-Y• cofactor, and self-assembly in vitro have been demonstrated [Cotruvo and Stubbe, 2010]. The enzyme has also shown to utilise a MnIIIMnIII-Y• cofactor in vivo under iron-limited conditions [Cotruvo and Stubbe, 2011a], and it has been established that E. coli requires manganese under conditions of
chronic H2O2 stress [Anjem, Varghese and Imlay, 2009]. Oxidative stress is another condition which is reported to increase the level of class Ib RNR expression [Cotruvo and Stubbe, 2011b]. Additionally, in two Bacillus species, the MnIIIMnIII-Y• form has been demonstrated to be more active than the iron forms [Crona et al., 2011].
1.2.2. Generation of an active metal cofactor – the importance of NrdH and NrdI The operon structures of the genes encoding the class I RNRs are nrdAB for subclass Ia, nrdHIEF for subclass Ib and nrdABPhe for subclass Ic. In the case of class Ia and Ic RNRs, the NrdA proteins corresponds to the R1 subunits and the NrdB/NrdBPhe proteins corresponds to the R2 subunits (Figure 1.2). In the case of class Ib RNRs, the NrdE protein corresponds to the R1 subunit and the NrdF protein corresponds to the R2 subunit. Two additional proteins, NrdH and NrdI, are required for the regulation of class Ib RNRs and is either cotranscribed with the nrdEF or often transcribed in a coordinated fashion when the operon structure is different, i.e. when nrdH and nrdI are located elsewhere in the genome [Jordan, Aragall et al., 1996; Monje-Casas et al., 2001; Torrents, Roca and Gibert, 2003].
NrdH is a 9 kDa protein with a conserved CXXC motif which is characteristic for thiol- disulfide oxidoreductases [Stehr et al., 2001]. Biochemical studies have demonstrated that it can act as an electron donor for NrdE, suggesting it to be the in vivo reducer of the two substrate-reducing cysteines in the active site (Figure 1.5) [Jordan et al., 1997; Jordan, Pontis et al., 1996]. In fact, NrdH have shown in vitro to be responsible for a 10-fold elevation of the activity of the manganese form of Bacillus anthracis and Bacillus cereus (B. cereus) class Ib RNRs compared to that of the iron form of the enzymes [Crona et al., 2011]. E. coli contains the genes for both class Ia and Ib RNR, and studies comparing the specificity and
physiological roles of NrdH culminates with NrdH being the physiological reducer of NrdE in this organism. NrdH receives electrons from the thioredoxin reductase that also delivers electrons to thioredoxin, which is the physiological reducer of NrdA in E. coli [Gon, Faulkner and Beckwith, 2006; Jordan et al., 1997]. NrdH may therefore be specific for class Ib RNR, and considered as an in vivo analogue to the thioredoxin or glutaredoxin in class Ia RNR (Figure 1.2), with which it has been shown to share considerable similarities [Jordan, Pontis et al., 1996, Stehr et al., 2001]. NrdH is, however, not considered a member of the
glutaredoxins because it lacks a specific sequence responsible for glutathione binding
[Bushweller et al., 1994]. In addition, the active site sequence C-M/V-Gln-C differs from the typical C-P-Y-C of glutaredoxins and from the conserved C-G-P-C of thioredoxins [Gon, Faulkner and Beckwith, 2006].
Figure 1.6: NrdI from B. cereus with a flavodoxin-like fold, where two α-helical layers sandwitches a 5-stranded parallel β-sheet. The flavin cofactor FMN (gray, on the top of the molecule) is in its oxidised form (PDB code 2X2O).
The figure was made using PyMOL (The PyMOL Molecular Graphics System, Version 1.5.0.4 Schrödinger, LLC).
Figure 1.7: Assembly of flavin based cofactors.
NrdI is a 15 kDa protein with a flavin mononucleotide (FMN) cofactor as the redox-active component and a flavodoxin-like fold (Figure 1.6 and Figure 1.7). The protein was isolated and characterised for the first time in 2008 [Roca et al., 2008; Johansson et al., 2010], when it was shown to be essential for the functionality of NrdHEF from Streptococcus pyogenes (S.
pyogenes).
FMN
Figure 1.8: Flavin redox states, using FMN as an example. The ox form is yellow, the sq form is blue and the hq form is colourless.
During the same year, the FMN cofactor of NrdI from E. coli was shown to possess unusual reduction-oxidation (redox) properties compared to those of typical flavodoxins (Flds). Flavin cofactors can have several different redox states; they can be in an oxidised (ox), partially reduced or fully reduced form. The partially reduced form can be anionic or neutral and is called semiquinone (sq). The fully reduced form can be anionic or neutral as well, and the neutral, fully reduced form is called hydroquinone (hq) (Figure 1.8).
Because of these redox properties, they are capable of switching between one- and two- electron transfer. On binding to a specific protein, the equilibrium between the different redox-states of the flavin can change dramatically. Some enzymes show essentially no stabilisation of sq, while others give almost 100 % stabilisation [Massey, 2000]. The
physiological role of typical Flds is to be a one-electron reductant, as the Eox/sq is between -50 and -220 mV and the Esq/hq is between -370 and - 450 mV. This means that the Eox/sq is
lowered compared to that of free FMN, whereas the Eox/sq is heightened [Cotruvo and Stubbe, 2008; Draper and Ingrahm, 1968; Hoover et al., 1999]. For NrdI, on the other hand, Eox/sq and Esq/hq are approximately equal, thereby making it a good candidate for a two-electron reducer (Table 1.1).
FMN free in solution Fld-bound FMN NrdI-bound FMN
Eox/sq (eV) -238 from -50 to -220 -264
Esq/hq (eV) -172 from -370 to -450 -255
Table 1.1: Reduction potentials of different redox-states of FMN free in solution, FMN bound to flavodoxin and FMN bound to the flavodoxin-like, NrdI, respectively.
This fundamental difference can be explained by the negative electrostatic environment around the FMN in typical Flds, which is replaced by a more neutral or positively charged environment in the case of NrdIs from Bacillus subtilis and E. coli, respectively. The negative electrostatic environment in typical Flds is proposed to hinder reduction of sq to hq, hence lowering Esq/hq [Bennett-Lovsey et al., 2008; Cotruvo and Stubbe, 2008; Johansson et al., 2010; Kelley, MacCallum and Sternberg, 2000; Zhou and Swenson, 1995]. Interestingly, a substitution of aspartate residue number 95 with an asparagine residue, thus converting this residue from a negative to a neutral one; in oxidised Fld from Desulfovibrio vulgaris increased Esq/hq by 46 mV. Because this residue, which is 6.3 Å from the flavin N(1), is
conserved as a D in most Flds and as an N in all NrdIs, it is proposed to play an important role in the destabilisation of the sq form of flavin mono dinucleotide (FMN) observed in E. coli NrdI [McCarthy et al., 2002; Cotruvo and Stubbe, 2008]. Moreover, the elevated Eox/sq level in NrdI may be explained by the presence of a flexible, glycine-rich loop in the vicinity of N(5) in of the FMN of most NrdIs, which is proposed to stabilise the ox form of flavin with its nonpolar and neutral properties [Cotruvo and Stubbe, 2008; Drennan et al., 1999; Hoover et al., 1999 and Ludwig et al., 1997].
Generation of an active cofactor in class Ib RNRs. A model of the mechanistic pathway for the generation of an active cofactor in class Ib RNRs has been developed [Cox et al, 2010;
Cotruvo and Stubbe, 2010]. The reaction mechanisms is based on a cofactor with manganese, and use either hydroperoxides (HO2-) or hydrogenperoxides (H2O2) to oxidise MnIIMnII and to extract an electron and a proton from tyrosine, resulting in the MnIIIMnIII-Y• cofactor (Figure 1.3b) [Cox et al, 2010; Cotruvo and Stubbe, 2010; Tomter et al., 2012]. Structural studies have revealed that upon complex formation between NrdI and NrdF, the active site channel in NrdF is extended towards the flavin ring in NrdI, defining a path from the FMN site in NrdI to the Mn site of NrdF. Moreover, a trapped species, which was best modelled as (the putative) peroxide, was observed in this channel. This finding supports the proposed mechanism of MnIIIMnIII-Y• assembly [Boal et al., 2010]. The fact that E. coli NrdI seems to have been fine- tuned for a rapid transfer of its second electron, and thereby ensures effective delivery of two electrons in two one-electron transfers [Cotruvo and Stubbe, 2008], further supports the proposed mechanism. Whether NrdI is present in stochiometric amounts relative to NrdF, or if an Fld reductase (FldR) allows NrdI to function catalytically, remains to be established.
However, the discovery of NrdI and its function have shed new light to a previous electron spin resonance study [Khangulov et al., 1990]. The study reveals that in vitro assembled
manganese-enzyme from Thermus thermophilus showed an activity level doubled compared to that of in vitro assembled iron-enzyme, and that the cofactor-to-protein ratio (Y•/β2) was approximately three times higher for the iron-enzyme. Viewed from the point of the proposed reaction mechanism, the finding that the class Ib RNR has a low Y• content, may result from a requirement of additional components in the system. An NrdI reductase could to facilitate the supply of peroxides for the mechanism and to re-reduce NrdI for multiple catalytic cycles [Cotruvo and Stubbe, 2011b].
1.3. Aims of the study
The focus of this master project has been on two FldRs encoded by B. cereus. Recent
discoveries have proposed that the radical generating component of the class Ib RNR system, i.e. the metal cofactor in the NrdF subunit, is a dimanganese-cluster instead of a diiron-cluster.
It has further been shown that the flavoprotein NrdI is essential for the generation of the tyrosyl radical when the metal is manganese. Hence, for a functional class Ib RNR system (with manganese), the NrdI itself needs to be reduced. Since NrdI is a flavodoxin-like protein, the most probable reductant for NrdI would be a FldR. Two FldRs have been identified in the genome of B. cereus, and the aim of this master project was to investigate them (Figure 1.9).
They are of the approximate same size, BC0385 is of 39 kD and BC4926 is of 36 kD, and the sequence identity is 41 %. Therefore, they are assumed to have similar functions in vivo. The goal was to be able to reveal if either FldRs could reduce NrdI, if a specific reductase is needed for this purpose and if one of these FldRs is that specific reductase. Such findings would enhance the knowledge about the function of NrdI, which still is unclear, and thus also tell more about the enzymatic process of ribonucleotide reduction in the class Ib RNR system, in B. cereus, specifically.
Figure 1.9: The class Ib RNR system in B. cereus. Two potential FldRs have been located in the genome of the species; FldR BC0385 and FldR BC4926 (red ring).
Potential reductases for NrdI
The investigation of the FldRs involves both their activities and their structures, and involves the following key goals:
Generation of FldR BC0385 expressing E. coli cells
Design primers to amplify BC0385 from B. cereus genomic DNA, clone the gene into a pET-22b expression vector, transform E. coli cells with the plasmid
Express and purify both FldRs
Growth, expression and purification of each FldR; develop a purification protocol for each FldR
Measure redox activity levels for both FldRs
Activity assays with each FldR; activity assays with each FldR and NrdI, respectively
Crystallise both FldRs; co-crystallise each FldR with NrdI
Screen for crystallisation condition(s) for each FldR, optimise the conditions, grow large crystals from these conditions and perform X-ray diffraction experiments on the crystals
Solve the crystal structures of FldR BC0385 and FldR BC4926
Use the data sets from the X-ray diffraction experiments to solve the crystal structures of the FldRs
1.4. Protein X-ray crystallography
One of the main aims of this project was to solve the 3D-structure of the two FldR candidates.
Protein crystallography is a powerful method that allows for determination of the crystal structure of proteins. Only the first steps of protein crystallography was used in this master project, and they are briefly introduced below.
1.4.1. Protein X-ray crystallography
Protein X-ray crystallography (PX) can be looked on as an analogue to a microscope. The maximal resolution the microscope gives, is limited by the wavelength of the radiation used.
A light microscope is e.g. based on the scattering of visual light, and cannot resolve details that are closer together than half the wavelength of visible light, which ranges from ̴ 400 nm to ̴ 700 nm [Lattman and Loll, 2008].
The distances between atoms in a molecule are typically 0.8-2.5 Å (0.08-0.25 nm; Figure 1.10). As X-rays have wavelengths between 0.5 Å and 1.6 Å, they can be used to visualise protein molecules and the sub-molecular details in them [Blow, 2002].
Figure 1.10: The spectrum of electromagnetic rays (light). Visible light ranges from ̴ 400 nm to ̴ 700 nm; X-rays ranges from ̴ 0.05 nm to ̴ 0.16 nm. The figure is taken from Wikipedia.
1.4.2. Scattering
Scattering, or diffraction, is the process by which matter deflects light rays from their straight lined path, without change of wavelength. In microscopy, light scattered by the specimen is captured and focused by lenses to make an image. In PX, the light (X-rays) scattered by the protein crystal is captured by a detector. An image of the protein is obtained by performing calculations that mimic the action of the lens [Lattman and Loll, 2008].
When an X-ray photon encounters an electron, it may be absorbed, and if so, the electron is set into vibration at the X-ray frequency. This vibrating electron then emits an X-ray photon of the original energy and wavelength [Blow, 2002].
1.4.3. The nature of protein crystals and crystal growth
To determine the structure of a protein molecule, the scattering effect from one single
molecule is not enough. Under the right conditions, protein molecules can pack together with a high degree of symmetry. In principle, each molecule of the three-dimentional protein crystal provides a copy of the same signal, thus enhancing the diffraction pattern and making it possible to visualise the protein molecule [Blow, 2002].
Crystallisation of proteins is based on bringing the protein solution into a supersaturated state (Figure 1.11). Factors such as pH and temperature, as well as concentration and type of salt and hydrophobic polymers, can affect protein solubility. The extents of the affection depend on the properties of the specific protein [Weber, 1997]. An example of a phase diagram of a protein is presented in Figure 1.11, where protein concentration is a function of salt
concentration. The curves define the different levels (zones) of supersaturation, which in turn determines at which protein and salt concentrations crystals are obtained. The level of
supersaturation also determines what kind of protein crystals that are grown [Drenth, 1999]. If the level of supersaturation is in the area of nucleation, protein molecules accumulates to form small protein aggregates. These aggregates serve as nuclei for crystal growth. To obtain a few, large crystals in stead of many small, it is necessary to achieve crystal nuclei in the nucleation zone, followed by transition to the growth zone. The growth zone possesses a lower degree of supersaturation, which supports crystal growth in stead of formation of more nuclei [Weber, 1997].
1.4.4. Vapour diffusion
The most frequently used crystallisation methods are the vapour diffusion techniques. The reservoir solution contains buffer, salt and eventually other additives. A droplet containing a mix of protein solution and reservoir solution is equilibrated against the reservoir solution in a defined, closed environment. The difference in precipitant concentration is driven towards equilibrium through vapour diffusion, which leads to a slow decrease of protein droplet volume and thus a slow increase of protein concentration in the droplet. Under the right conditions, the concentration is increased into the nucleation zone and thereafter decreased into the growth zone, and a few, large crystals are obtained. Two types of vapour diffusion setups are presented in Figure 1.12; namely hanging drop and sitting drop.
Figure 1.11: Typical solubility curves in the phase diagram of a protein. Crystal nucleation is a critical phenomenon that occurs only in the supersaturation zone. Crystals grow under conditions of supersaturation after nuclei have taken form. A typical path for obtaining a few, large crystals, is presented as red arrows.
Figure 1.12: Two types of vapour diffusion setup.
2. Methods
Presented here are the methods used to study the two proteins of interest, i.e. the two (potential) FldRs – BC0385 and BC4926 – which are encoded in the genome of B. cereus.
First, it is described how the generation of FldR BC0385-expressing cells was performed, which included the polymerase chain reaction (PCR) for amplification of the BC0385 gene, restriction cutting of the insert (the gene) and the vector with subsequent ligation of the two DNA molecules, and transformation of competent E. coli cells with the ligation product.
Thereafter, a standard procedure was developed for the expression and purification of the two proteins, followed by procedures for crystallisation of the proteins, X-ray diffraction
experiments and measurement of protein activity.
2.1. Cloning of BC0385 into the pET-22b expression vector and generation of FldR BC0385 expressing E. coli cells
In this project, the expression vector pET-22b was utilised to facilitate (over)expression of both FldRs. In the pET system, target genes are expressed from the plasmid under control of T7 polymerase transcription. Transcription of both T7 polymerase and the gene of interest is activated by the presence of Isopropyl-β-D-Thiogalactopyranoside (IPTG) [Novagen, 2010].
For one of the FldRs it had previously been developed an expression system with the pET-22b vector and E. coli BL21 (DE3) cells, and a similar expression system had to be developed for the other FldR.
2.1.1. Amplification of BC0385
PCR was performed to generate BC0385 DNA fragments for subsequent cloning into the pET-22b expression vector. The DNA sequence of the pET-22b vector contains restriction sites for a number of restriction enzymes; in the process of amplifying a gene by the use of PCR, it is possible to introduce the same restriction sites flanking the gene which is subjected
to cloning. This is performed by designing primers containing these restriction sites, which is then used in the PCR.
A small sample of the provided stock of B. cereus ATCC 14579 was spread onto agar plates (Table A4.2) in a sterile environment and incubated at 37 °C for 15 hours. A 3 ml Luria Broth (LB) culture was made from one of the largest single colonies and incubated at 37 °C with 250 rpm shaking until a 1:4-dilution of the culture had reached OD600 = 0.90. Plasmid was purified from the culture using the NucleoSpin plasmid kit with a slightly modified protocol (spinning down 3 µL of cell culture instead of 1 µL; eluate with 35 µL instead of 50 µL in the final step). The concentration of B. cereus genomic DNA in the sample was measured using NanoDrop (ND-1000, Nanodrop Technologies).
Forward and reverse primers with NdeI and HindIII restriction sites were designed for BC0385 (section A5.3.).
PCR was performed with the primers designed for BC0385 and B. cereus genomic DNA as template using a gradient for the annealing temperature (Table A2.1). The size of the PCR product from the eight different reaction mixtures was tested and purified by agarose gel electrophoresis with a 0.8 % gel (Table A4.1). The bands containing DNA fragments of the size corresponding to the length of BC0385 (1050 base pairs) were excised from the gel and further purified using the Wizard SV Gel and PCR Clean-Up System kit. The concentration of PCR product in the sample was measured using NanoDrop.
2.1.2. Purification of pET-22b vector
Amplification and purification of pET-22b expression vector was performed, in order to generate a sample with a vector concentration suitable for the process of cloning the BC0385 into the vector.
A small sample of the provided stock of E. coli bacterial strain TOP10 containing the pET- 22b vector was spread onto agar plates (100 µg/mL ampicillin) and incubated at 37 °C for 15 hours. 6 parallels of 6 mL LB cultures (100 µg/mL ampicillin) were made from each of the 6 largest single colonies and incubated at 37 °C and 250 rpm for 17 hours. Plasmid was purified from the cultures using the NucleoSpin plasmid kit, with a somewhat modified protocol due
to initial difficulties in getting a high enough concentration of plasmid in the sample (section A.6.1). The concentration of pET-22b vector in the sample was measured using NanoDrop.
2.1.3. Restriction cutting and ligation of BC0385 and pET-22b vector
The plasmid and the DNA fragment of the gene to be cloned are cut with the same two restriction enzymes to introduce compatible ends in the respective molecules, which
subsequently can be ligated together. By using two different restriction enzymes in stead of one, it is made sure that the gene (or DNA fragment) is inserted into the vector in the appropriate direction.
The PCR product from the amplification of BC0385 and the prepared pET-22b vector were separately double digested with restriction enzymes NdeI and HindIII and incubated in a 37
°C water bath for 1 hour (Table A3.1). The FastDigest buffers and enzyme solutions were used since they turned out to give positive results (Table A3.1). After the normal (i.e. not Fast-) buffers and enzyme solutions was tried out with different incubation times and
temperatures, but gave only negative results. The digestion products were purified by agarose gel electrophoresis with a 0.8 % gel (Table A4.1). The bands containing DNA fragments of the size corresponding to the length of pET-22b (5493 base pairs) and BC0385 (1050 + 4 from primers = 1054 base pairs) were excised from the gel and further purified using the Wizard SV Gel and PCR Clean-Up System kit. The concentrations of PCR-product and pET- 22b vector in the samples were measured using NanoDrop.
The purified digestion products were ligated with T4 ligase and a 3 times molar excess of BC0385 in room temperature for 1 hour (Table A3.2).
A small amount of the ligation product was analysed by agarose gel electrophoresis to ensure that the plasmid corresponded to the length of pET-22b with the BC0385 insert (6547 base pairs).
2.1.4. Transformation of competent hosts with the ligation product
E. coli XL10-Gold ultracompetent cells are created for transformation of large DNA molecules with high efficiency. These cells were transformed with the ligation product as follows.
1) Thaw the cells on wet ice for 5 minutes
2) Add 5 µL of ligation mixture and incubate on ice for 30 minutes 3) Heat shock the cells in a 42 °C water bath for 1 minute
4) Cool down the cells on wet ice for 2 minutes
5) Add 900 µL SOC medium which are pre-heated to 42 °C and incubate in a 37 °C water bath for 30 minutes.
A small sample of the transformation mixture was spread onto agar plates (100 µg/mL ampicillin) and incubated at 37 °C for 15 hours. A 3 mL LB culture (100 µg/mL ampicillin) was made from one of the largest single colonies and incubated at 37 °C and 250 rpm for 17 hours. Plasmid was purified from the culture using the NucleoSpin plasmid kit with the same modifications of the protocol as before. The concentration of plasmid in the sample was measured with NanoDrop.
The plasmid was sequenced in three parallels at the ABI-lab at the University of Oslo (sample details in Appendix 5).
2.1.5. Site-directed mutagenesis
To back-mutate a so-called missense mutation – i.e. a mutation leading to amino acid
substitution in the gene product – that had occurred in the BC0385 sequence during PCR, the vector with the mutated insert was subjected to design of mutagenic primers and subsequent PCR with these.
Forward and reverse primers mutagenic for the substituted base pair number 114 in BC0385 were designed and ordered (section A5.4.), and PCR was performed with the primers and the ligation product as template (Table A2.2).
DpnI endonuclease has the target sequence 5’-Gm6ATC-3’ and is used to remove template DNA. Because the newly synthesised DNA molecules from the PCR are not methylated, this is an efficient method to remove the template, which is not wanted to be a participant in cell transformation [Weiner et al., 1994].
To destroy the template, 1 µL of DpnI (10 u/µL) was added directly to the PCR tubes after the end of the PCR program and the mixture was incubated for 1 hour at 37 °C. A small amount of the PCR product was analysed by agarose gel electrophoresis with a 0.8 % gel (Table A4.1) to ensure that the DNA was of the right size (6547 base pairs).
E. coli XL10-Gold ultracompentent cells were transformed with the PCR product after the same protocol as for the ligation product (section 2.1.4.), but in step 2) 10 µL PCR mixture was added instead of 5 µL ligation mixture.
A small sample of the transformation mixture was spread onto agar plates (100 µg/mL ampicillin) and incubated at 37 °C for 15 hours. A 3 mL culture (100 µg/mL ampicillin) was made from one of the largest single colonies and incubated at 37 °C and 250 rpm for 17 hours. Plasmid was purified from the culture using the NucleoSpin plasmid kit with the same modifications of the protocol as before. The concentration of plasmid in the sample was measured with NanoDrop.
The plasmid was sequenced at the ABI-lab at the University of Oslo in six parallels; three using the T7 polymerase and three using the T7 terminator polymerase (sample details in section A.5.1.).
2.1.6. Transformation of expression hosts with the plasmid
E. coli BL21 (DE3) competent cells are developed with regard to effective protein expression.
E. coli BL21 (DE3) competent cells were transformed with the plasmid (i.e. the pET-22b vector with inserted BC0385) after the same protocol as for the ligation product (section 2.4.1.), but in step 2, 3 µL plasmid sample was added instead of 5 µL ligation mixture.
A small sample of the transformation mixture was spread onto agar plates containing 100 µg/mL ampicillin and incubated at 37 °C for 15 hours. A 200 mL LB culture (100 µg/mL ampicillin) was made from one of the largest single colonies and incubated at 37 °C and 250
rpm for 17 hours. 4 mL LB culture was diluted in a new 200 mL LB culture (100 µg/mL ampicillin) and incubated for 7 hours at 37 °C. Stocks were made by mixing 500 µL of LB cultured cells with 500 µL glycerol (60 %) and stored at -80 °C.
2.2. Expression and purification of FldR BC0385 and FldR BC4926
A stock of E. coli cells of the strain BL21 (DE3), containing a vector with the BC4926 gene inserted, was previously generated, and kindly provided, by Åsmund K. Røhr. Possessing E.
coli cells expressing both FldRs, respectively, the next step was to produce and purify the two proteins for subsequent crystallisation experiments and activity measurements.
The following procedure of expression and purification applies for both FldR BC0385 and FldR BC4926, if not specified otherwise.
2.2.1. Growth of FldR expressing E. coli cells
A small sample of the freeze stock of FldR-expressing E. coli was spread onto agar plates (100 µg/mL ampicillin) and incubated at 37 °C for 15 hours. A 200 mL LB culture (100 µg/mL ampicillin) was made from one of the largest single colonies and incubated at 37 °C and with 250 rpm shaking for 17 hours. 4 mL of this LB culture was diluted in a new 200 mL LB culture (100 µg/mL ampicillin) and incubated for 7 hours at 37 °C. 6 x 10 mL samples of the new LB culture was then diluted in 6 x 1000 mL Terrific Broth (TB) cultures (100 µg/mL ampicillin) and incubated at 37 °C and 250 rpm until OD600 = 0.9 ± 0.1 (for about 2-3 hours).
The TB cultures were then put on ice until the temperature had dropped to 20 °C. A 50 µL sample was taken from one of the TB cultures for an expression test (section 2.2.2.), centrifuged for 1 min at 19,000 g and stored at -80 °C. Over-expression of FldR in the 6 cultures was induced with 1 mM IPTG and they were incubated at 20 °C and 250 rpm for 14 hours. A 25 µL sample was taken from one of the cultures for the expression test (see below), centrifuged for 1 min at 19,000 g and stored at -80 °C. Bacteria was harvested at 5000 g for 15 minutes, and the bacteria paste was divided into samples of approximately 15 g and stored in zip lock plastic bags at -20 °C.
2.2.2. Testing the expression of FldR
The protein content of the TB culture before and after induction of over expression (and subsequent growth for 15 hours) was analysed by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). This was to investigate if the E. coli cells actually expressed the FldRs.
The pellets from the 50 µL and 25 µL TB culture samples were suspended in 50 µL and 100 µL protein cracking solution (Table A4.3), respectively. The samples were vortexed, heated to 90 °C for 2 minutes and vortexed again. 1 µL of each mix was applied to a 10-15 % gradient gel for SDS-PAGE, and subject to a pre-programmed separation script on the PhastGel system, specific for the gradient in the gel.
2.2.3. Lysis of FldR expressing E. coli cells
The cell walls of bacteria have a rigid layer, called peptidoglycan, which is primarily responsible for the strength of the wall. In gram-negative bacteria, such as E. coli, an additional layer is present outside the rigid layer, called the outer membrane or the
lipopolysaccharide layer. Due to this rigidity, a high pressure is required to rupture the cell wall of E. coli cells [Mardigan and Martinko, 2006]. The principle of the X-press® is to break frozen cells by squeezing them through a small opening under high pressure, which will disrupt the cell wall and membrane [Magnusson and Edelbo, 1976].
Approximately 15 g of bacteria paste was cooled down with N2 (l) and thoroughly crushed before applied to the X-press equipment, which had been pre-cooled to -20 °C. The
equipment was put together and the bacteria were lysed by being pushed back and forth through a tiny hole by the use of an air compressor.
From this point and through the rest of the purification process, it was made sure that the protein solution was kept at 4 °C by working in a 4 °C room or on ice, pre-chilling solutions before applying them to the protein solution as well as using a centrifuge cooled to 4 °C. The protein solution was stored at -20 °C during longer halts in the purification procedures.
The lysed bacteria were dissolved in 100 mL buffer A (Table A4.4) over a time period of 30 minutes, and thereafter centrifuged at 39,000 g for 1 hour. The supernatant was kept for the next step.
2.2.4. Purification of FldR: DNA precipitation
Added in small increments to solutions of low ionic strength, streptomycin (first) precipitates DNA [Cohen and Lichtenstein, 1960]. Therefore, as the first step in the protein purification process, Streptomycin sulphate was used to remove DNA from the solution.
A 10 % streptomycin sulphate solution corresponding to ¼ of the supernatant was added over a period of 5 minutes while stirring at low speed on a magnetic stirrer, before subsequent stirring at low speed for another 10 minutes. The solution was centrifuged at 39,000 g for 30 minutes and the supernatant was kept for the next step.
2.2.5. Purification of FldR: Protein precipitation by salting out
At high ionic strengths, the solubility of proteins decreases. When the salt concentration in a protein solution is increased, the solvent is becoming occupied in polar interactions with the ions, i.e. solvating the salt ions. The reduced amount of solvent available to engage in interactions with the protein, i.e. solvate the protein, causes the protein to precipitate. The precipitation of proteins caused by the formation of hydrophobic interactions with each other is often executed by the use of ammonium sulphate ((NH4)2(SO)4). The salt concentration at which a protein precipitates differs from one protein to another. Hence, salting out can be used to fractionate proteins [Voet and Voet, 2004].
A specified amount of (NH4)2(SO)4 salt per mL supernatant was added over a period of 15 minutes on a magnetic stirrer and stirred at low speed for another 20 minutes; 0.43 g/mL for the FldR BC0385 and 0.22 g/mL for the FldR BC4929. After precipitation the solution was centrifuged at 39,000 g for 20 minutes, and the protein pellet was kept for the next step.
This purification step had previously been optimised with respect to FldR BC4926 showing that 0.22 g (NH4)2(SO)4 per mL supernatant was enough to salt out all the FldR BC4926 protein from solution. With respect to the FldR BC0385, the required amount of (NH4)2(SO)4
was not optimised because of time limitations; a standard amount of 0.43 g/mL was chosen.
From this point and through the rest of the purification process, it was made sure that no bubbles were generated in the protein solution, as this could lead to protein denaturation.
2.2.6. Purification of FldR: Desalting chromatography
Prior to anion exchange chromatography, desalting chromatography (low resolution size exclusion chromatography) of the protein sample was performed in order to lower the salt concentration (see section 2.2.9. for principle). Proteins in general absorb light at 280 nm [Voet and Voet, 2004], and this property was used to monitor the proteins in this purification step.
The pellet was dissolved in 10 mL Buffer A (Table A4.4.) over a period of 1-2 hours, with gentle stirring and pipetting now and then. The protein solution was filtered through a 0.22 µm filter before subsequent desalting on a 2 x 5 mL HiTrap desalting column on the Äkta purifier system. The sample was eluted with Buffer A (Table A4.4), and the fractions showing absorption at 280 nm and low conductivity were collected for further purification.
2.2.7. Purification of FldR: anion exchange chromatography
Ion exchange chromatography is a method used to separate proteins according to differences in charge and charge distribution. The separation is based on the reversible adsorption of charged protein molecules to immobilised groups in the column material, and these groups have a charge opposite to that of the protein molecules. When negatively charged proteins are adsorbed to a positively charged column material, they can be released by gradually
increasing the concentration of salt (e.g. potassium chloride) in the elution buffer. Then the chloride ions will compete with the proteins in binding to the positively charged column material, and the proteins with a low density of net negative charge will tend to emerge first, followed by those having a higher negative charge density [Berg, Tymoczko and Stryer, 2007]. Proteins in general absorb light at 280 nm, and flavins absorb light at 450 nm in their oxidised state [Massey, 2000], and these properties were used to monitor the flavin-containing FldR proteins in this purification step.
Protein fractions from desalting chromatography were filtered through a 0.22 µm filter before subsequent purification on a Q HP anion exchange column. The protein was eluted with buffer B (Table A4.4), and by increasing the salt concentration with a linear gradient using buffer C (Table A4.4). The fractions showing absorption at both 280 nm and 450 nm were collected for further purification.
As a part of developing a standard procedure, the pH in the protein sample and in the ion exchange buffers (Buffer B and Buffer C) was increased until the protein was negatively charged and hence adsorbed to the column material.
To further optimise the standard procedure, the salt gradient was adjusted to go from 0 % and to an upper value defined by the point where all the protein of interest had been eluted. A part of this was also to choose a gradient length which gave a good separation without being too time consuming.
The collected fractions were analysed by SDS-PAGE to confirm the size of the protein in the collected peak(s) to see if it corresponded to the calculated size of the FldR. A small sample of the protein solution was mixed 50:50 with protein cracking solution, and the rest of the procedure is previously described in section 2.2.2. The fractions containing the protein of interest were pooled and kept for further purification.
Reconstitution of FldR BC4926 from apo protein and flavin
Before further purification, FldR BC4926 was tried reconstituted from apo-FldR and flavin.
First, to determine if apo-FldR binds either FMN or FAD, a small sample (2 mL) of apo-FldR from anion exchange chromatography was divided in two equal samples (1 mL); these were each mixed with one of the two flavin cofactors, FAD and FMN, respectively, to a
concentration of 1 mM. Each sample was filtered through a 0.22 µm filter before subsequent purification on a 2 x 5 mL HiTrap desalting column to determine the amount of cofactor that bound to the FldR.
The flavin cofactor which appeared to bind most effectively was used to reconstitute higher quantities of protein, but this time with cofactor to a concentration of 0.1 mM. The
reconstituted protein was purified by gel filtration chromatography (section 2.2.9.).
2.2.9. Purification of FldR: Gel filtration chromatography
Gel filtration chromatography is, like desalting chromatography, size exclusion, but with a higher resolution. The column consists of porous beads made of a polymer; small molecules can enter these beads, but large ones can not. The result is that small molecules are distributed
in the aqueous solution both inside the beads and between them, whereas large molecules are only located in the solution between the beads. Hence, large proteins are eluted from the gel filtration column before smaller proteins [Berg, Tymoczko and Stryer, 2007].
The collected protein solution from the anion exchange chromatography was concentrated using an Amicon ultrafiltration cell containing a filter with a cut-off value of 30,000 Da. Due to precipitation of the FldR and the reconstituted FldR, the protein solution was concentrated until protein almost started to precipitate. Samples of 25-50 µL were applied to a Sephadex G75 gel filtration column eluted with Buffer D (Table A4.4). Fractions absorbing light at 450 nm were collected.
2.3. Crystallization of FldR BC0385 and FldR BC4926
Protein crystallisation can be thought of as a two-stage process; 1) identify one or more conditions in which a certain protein precipitates as ordered crystals and 2) optimise the condition(s) to produce protein crystals of a quality that makes them suitable for X-ray diffraction experiments. To identify conditions in which a protein crystallises, different
buffers, pH values, precipitants, temperatures and crystallisation methods are usually varied in a systematic or a random way, referred to as screening. The different parameters usually affect the outcome to a large extent, both when it comes to if crystals are obtained or not, and when it comes to the shape and quality of the crystals that are eventually obtained.
2.3.1. Screening for crystallisation conditions
The vapour diffusion technique with a sitting drop setup was used to screen for crystallisation conditions. All the screening was performed at the Oryx 6 Robot (kindly provided by the group of Magnar Bjørås at OUS Rikshospitalet, Oslo, Norway) by the use of different commercial crystallisation screens (Table A1.7). The screens have typically 96 different conditions with varying precipitants, pH and additives. Crystallisation trays with two drops for each condition were used; concentrated protein was deposited as the first drop while the second drop normally had lower protein concentration. The different screens used in the
screening experiments with the different FldRs are presented in Table 2.3.1, as well as the content of protein and eventual other diluents in the second drop.
Normally, the protein solution after gel filtration was concentrated until protein precipitation was barely indicated. The protein solution was then centrifuged at 19,000 g for one minute and precipitated protein was discarded before the screening was performed. In the case of apo-FldR BC4926, the protein solution after gel filtration was concentrated to ̴ 15 mg/mL in one screening experiment and to 60 mg/mL in a separate screening experiment.
Table 2.3.1: The different screens and the content of the second drop in the screening experiments with the different FldRs.
Protein Screens Content of second drop
FldR BC0385 JSCG+ and Index 50 % protein FldR BC4926,
reconstituted
JSCG+, Natrix (E-H) and PEG/Ion(A-D) and Wizard I+II
100 % protein and a small excess of FAD
apo-FldR BC4926
(̴ 15 mg/mL) JSCG+, PEG/Ion and Wizard I+II 50 % protein apo-FldR BC4926
(̴ 60 mg/mL) JSCG+ and Midas 50 % protein FldR BC4926 Index, JSCG+ and Wizard I+II Buffer*
* A second drop was not set up because of limited amounts of protein solution.
The conditions resulting in hits (e.g. small crystals) were tried optimised for the growth of large crystals by varying the protein and precipitant concentration in a systematic way.
2.3.2. Optimisation of crystallisation conditions for apo-FldR BC4926
The condition in well H11 in the Wizard I+II screen (details in Table A.4.5) was tried optimised for the most diluted sample of FldR BC4926 apo form. The concentrations of zinc acetate (ZnAc2) and protein were both varied between 25 % and 100 %, in separate
experiments.
The conditions in wells H10 and H11 in the Wizard I+II screen (details in Table A.4.5) was further investigated for the 15 mg/mL sample of FldR BC4926 apo form. The concentrations of polyethylene glycol (PEG) and protein were varied separately between 25 % and 100 %.
This was performed in two parallels for both the H10 and H11 conditions; one parallel at room temperature and one parallel at 4 °C. In addition, two experiments with 50 % and 25 % protein were set up for the H10 condition, respectively, both with PEG concentration reduced to 25 %.
2.3.3. Optimisation of crystallization conditions for FldR BC4926
The conditions in wells A7 and B3 in the JSCG+ (details in Table A.4.6) screen was further investigated. The concentrations of PEG and protein were varied separately at 4 °C.
2.4. X-ray diffraction, data collection and data processing
Prior to X-ray diffraction experiments each of the crystals were transferred into a cryo solution containing the crystallization solution (70%) mixed with glycerol (30 %) as antifreeze before subsequent freezing in liquid nitrogen.
The X-ray diffraction experiments were performed at beam line ID29 at the European Synchrotron Radiation Facility (ESRF), Grenoble, France and at beam line X10SA at the Swiss-Light Source (SLS), at Paul Scherrer Institut (PSI), Villigen, Switzerland.
The diffraction images from the initial experiments were indexed with MOSFLM and the space group determined with POINTLESS [Leslie and Powell, 2007; Evans, 2005].
2.5. Activity measurements
2.5.1. Determination of the specific activity of the FldRs with DCPIP
In order to compare the activity of the two FldRs, the ability of each FldR to catalyse the reduction of 2,6-dichlorophenolindophenol (DCPIP) with NADH as electron donor was determined. The reduction of DCPIP can be measured spectroscopically; it has a light absorption maximum at 600 nm wavelength in its oxidised state, but in its reduced state it exhibits no absorption at this wavelength [Armstrong, 1965]. Thus, the electron transfer from NADH via FldR to DCPIP can be determined by measuring the drop in absorption at 600 nm (A600). The setup and the procedure were based on standard procedures by Worthington Biochemical Corporation (2012).
The initial velocity of DCPIP reduction was measured at different DCPIP concentrations as follows. Because the reduction of DCPIP by NADH takes place also in the absence of a catalyst [Dupuy, 2004], two parallels were measured for each reaction, one with FldR and one without. During the calculations, the amount of DCPIP reduced in the absence of FldR was subtracted from the amount that was reduced in the presence of FldR.
1) Mix Tris·HCl buffer (200 mM, pH 8.0) and DCPIP (different concentrations, ranging from 4 µM to 800 µM) in an eppendorf tube
2) Start the timer at the moment NADH (300 µM) is added to the tube 3) Add water (for blind reaction) or FldR (0.42 µM) to the tube
4) Mix the components by pipetting up and down one time, and transfer the solution to a cuvette
5) Measure the light absorption of the solution for the first time when the timer shows 30 seconds, and measure a new absorption spectrum every 15 seconds during 240 seconds 6) Repeat the procedure with a different concentration of DCPIP
A705 and A800 was measured in addition to A600. For a specific A600 around 1 absorbance unit (AU), it was observed that A705 was 1/5 of this value. This knowledge was used when the concentration of DCPIP gave absorption values exceeding 1.5 AU; the A705 was multiplied with 5 and reported as A600. DCPIP does not absorb any light at 800 nm. By measuring A800, it was ensured that the different spectra had the same base line.
The initial velocity (V0) of each reaction was determined (by “initial velocity” it is meant the velocity in the time range which the reaction was linear). The velocity was plotted against the concentration of DCPIP for each reaction in the series of different concentrations, and Origin (Origin 8.6, OriginLab Corporation) was used to estimate a Michaelis-Menten curve fit for the plot.
Before the experiment was carried out, the effects of different buffer concentrations and pH values were investigated, as well as the presence and absence of zinc chloride (ZnCl2), which was a part of the original setup. This was done in order to optimise the reaction conditions with regard to the functionality of the FldRs.
2.5.2. Comparison of the FldR activities with cytochrome c
In order to compare the reducing effect each FldR has on a different substrate, the reduction of cytochrome c (Cytc) with NADPH as the electron donor was tested after inspiration from Leadbeater et al. (2000). The reduction can be determined by measuring the increase in light absorption at 550 nm, as only Cytc in its reduced form exhibits absorption at this wavelength [Krab and Wikström, 1979].
Phosphate buffer (10 mM, pH 8.0), CytC (7 µM), NADPH (3 mM) and FldR (0.17 µM) were mixed together, and absorption spectra were measured after 0, 90, 180, 270 and 360 seconds.
2.5.3. Testing of the ability of the FldRs to reduce flavodoxins from B. cereus
FMN absorbs light at 460 nm in its oxidised form. The absorption peak at this wavelength is decreased when FMN is in the sq (partially reduced) form, and a new peak appears at 600 nm.
When the cofactor is in the hq (fully reduced) form, both these peaks have disappeared [Berg, Tymoczko and Stryer, 2007]. These properties of the FMN cofactor were used to monitor the reduction of NrdI.
The ability of the FldRs to reduce the flavodoxin-like protein NrdI from B. cereus was tested in the following manner. (The NrdI was previously purified in our group.)
1) Tris·HCl buffer (100 mM; pH 7.5) and NrdI (10 µM) was mixed together in a cuvette and made anaerobic with a Schlenk line, before the first absorption spectrum was measured.
2) NADPH (100 µM) was made anaerobic and added to the mixture, before the second absorption spectrum was measured.
3) FldR (1.6 µM) was added to the mixture, and absorption spectra was measured after 20 seconds and 30 minutes.
3. Results and discussion
3.1. Cloning of BC0385 into the pET-22b expression vector and generation of FldR BC0385 expressing E. coli cells
To generate and amplify BC0385 gene fragments for later cloning into a pET-22b vector, PCR was performed with primers designed with NdeI and HindIII restriction sites and B.
cereus genomic DNA as template. The PCR mixtures from each of the wells in the
temperature gradient (Table A2.1) were tested and purified with agarose gel electrophoresis (Figure 3.1.1). The picture of the gel taken with an ultraviolet light-camera showed that all annealing temperatures tested turned out to give a similar amount of PCR product. It was therefore not possible to determine which of the temperatures in the selected range that was the most optimal. However, measurement of the DNA concentrations in the respective reactions after purification from the agarose gel indicated that the PCR with annealing temperatures 57.6 °C and 66.3 °C were the ones with the highest concentrations; 57.1 ng/µL and 60.8 ng/ µL, respectively. It can be derived from this that these two temperatures were the most effective for the primers that were designed and used (section A5.3). Therefore, these two samples of BC0385 were chosen for further work.
Figure 3.1.1: PCR mixtures in agarose gel after reaction and separation by electrophoresis. A + E) DNA standard (Figure A1.1), B-D + F-I) PCR mixtures with annealing temperatures increasing from left to right.
